Research map / Detection and segmentation / Open-set and few-shot detection
Planning with Learned Object Importance in Large Problem Instances using Graph Neural Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Predict and plan with a learned sufficient subset of objects rather than all objects | PLOI versus pure planning · Gripper | planning time (seconds): 24.48 -> 0.47 | Helped |
| Predict and plan with a learned sufficient subset of objects rather than all objects | PLOI versus pure planning · Miconic | planning time (seconds): 87.71 -> 21.64 | Helped |
| Predict and plan with a learned sufficient subset of objects rather than all objects | PLOI versus pure planning · Hanoi | planning time (seconds): 3.19 -> 3.39 | No clear effect |
| Replace the trained object scorer with uniformly random importance scores (ablation) | PLOI · Gripper | planning time (seconds): 0.47 -> 56.58 | Hurt |
| Replace the trained object scorer with uniformly random importance scores (ablation) | PLOI · Hanoi | planning time (seconds): 3.39 -> 3.47 | No clear effect |
| Use one message-passing iteration instead of the selected setting (ablation) | PLOI on Blocks · Blocks | planning performance | Hurt |
| Vary the number of training problems; performance peaks starting at 3 and remains robust above 3 (ablation) | PLOI on Blocks · Blocks | planning performance | Mixed |
| Add L2 regularization, dropout, or hyperparameter search (ablation) | PLOI GNN · All evaluated methods | planning performance | No clear effect |
| Use object-level importance prediction instead of action grounding | PLOI versus ILP AG · Gripper | planning time (seconds): 5.71 -> 0.47 | Helped |
| Use object-level importance prediction instead of action grounding | PLOI versus GNN AG · Blocks | planning time (seconds): 52.95 -> 0.62 | Helped |
| Use object-level importance prediction instead of a reactive policy | PLOI versus Policy · Gripper | planning time (seconds): 28.94 -> 0.47 | Helped |
| Use learned object selection with incremental planning and validation | PLOI versus Pure planning · Triangle Tireworld | failure rate: 0.01 -> 0.08 | Hurt |
Closest papers by meaning
- Graph Learning for Numeric Planning (NeurIPS 2024 · similarity 0.88)
- Learning Domain-Independent Heuristics for Grounded and Lifted Planning (AAAI 2024 · similarity 0.87)
- Graph Neural Network Based Action Ranking for Planning (NeurIPS 2025 · similarity 0.87)
- State Encodings for GNN-Based Lifted Planners (AAAI 2025 · similarity 0.86)
- Efficient Lookahead Encoding and Abstracted Width for Learning General Policies in Classical Planning (NeurIPS 2026 · similarity 0.86)
- System 1.x: Learning to Balance Fast and Slow Planning with Language Models (ICLR 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.